IGComments

Guides

Using exported Instagram comments in research: method notes

An export is a convenient way to get a post’s public comments into a table. It is not a neutral census. Here are the points a methods section should state, and the limits a reviewer is likely to ask about.

Updated Nov 17, 2026 · 8 min read

What the data is

For each public comment and reply that Instagram serves, an export gives: the username, the comment text, a UTC timestamp, the like count, whether it is a reply, the username and comment ID of the parent for replies, mentions and hashtags extracted from the text, the comment ID and the user ID. Formats are CSV, Excel and JSON. The JSON file also records the post URL, the account that posted it, the comment count Instagram reported at the time (comment_count_reported) and the time of the fetch (fetched_at). Those fields are what make a dataset describable.

It does not contain profile data (names, bios, follower counts, locations), images, or anything from private accounts, Stories or Lives.

Record the snapshot

A post is a living object. Comments are added, deleted and moderated, likes change, and the text of a comment can be altered or removed. An export is a snapshot of the public state at the moment of fetching. In your methods, give:

  • the post URL and account, and the date and time of fetch in UTC (from fetched_at in the JSON);
  • the comment count Instagram displayed at that time and the number of rows in your file;
  • how you chose the post (and your sampling logic if you used several);
  • the tool and its date of use. The export stays downloadable for 30 days, so keep your own copy of the original.

Store the untouched file, compute a checksum (shasum -a 256 file.csv on a Mac or Linux), and do all cleaning in a copy.

What is not retrievable, stated plainly

Instagram’s displayed comment count includes comments it does not serve to anyone: spam-filtered comments, comments caught by an account’s hidden-words settings, comments from restricted accounts and comments from deleted accounts. On large giveaway posts a complete export is typically about 85 to 92 percent of the displayed count, and the results page shows the comments read next to the count Instagram reports. Treat the shortfall as non-random missingness: the missing comments are likelier to be spam or abusive, so analyses of toxicity, spam or hostility based on public comments probably understate it. State this in the limitations. The mechanics are in why the count does not match.

Other limits to name: only public posts, no stories or lives, replies are one level deep, edits are not tracked, and a single export gives one moment in time rather than a history.

Sampling and representativeness

Commenters are a self-selected subset of viewers, and viewers are a subset of followers. Comments from one post tell you about that post’s commenters, not about the account’s audience, nor about Instagram users in general. If you generalise, choose posts by a stated rule (all posts in a date range, a random sample of posts) and describe it. Check concentration before you treat comments as independent: count comments per user_id, because a handful of accounts can account for a large share of a post. In Excel 365 or Google Sheets, =COUNTA(UNIQUE(D2:D5000)) gives distinct accounts from the user IDs.

Use user_id, not the username, to identify accounts across datasets: usernames can change, and the ID stays the same.

Ethics, privacy and the law: pointers

Public does not mean free of obligations. Many ethics boards and data protection regimes treat usernames and comment text as personal data even when public. Points to raise with your review board or data protection officer before you start:

  • Minimise. Drop columns you do not need. Pseudonymise usernames and user IDs in the working copy and keep the key separately and securely.
  • Quoting. A verbatim comment can be found by searching for it, which re-identifies the commenter. Paraphrase or get consent if participants could be harmed.
  • Sensitive topics. Comments about health, politics or identity need a higher bar.
  • Sharing. Check Instagram’s terms and your institution’s rules before you deposit a dataset in a repository. Many researchers share IDs or derived measures instead of text.
  • Retention. Set a delete date for the raw file.

For the legal side, which depends on where you are and what you do with the data, see is it legal to export Instagram comments. It is general information, not legal advice, and your institution’s counsel is the authority.

Citing the dataset

A workable pattern, adapted to your style guide: Author or account. (Year). Comments on an Instagram post [dataset]. Collected on [date, UTC] with IGComments (igcomments.com). Post: [URL]. N = [rows] of [displayed count] displayed comments. Put the dataset description in an appendix, with the column list, the exclusions and the checksum. If you use the tool for a large study, the business page covers invoices for grants and institutions, and very large posts are collected in the background over one to three days.

To start, paste a post into the tool; up to 200 comments are free.

FAQ

Does the export include everything Instagram stores about a post?

No. It includes the public comments and replies Instagram serves. Comments that are filtered, hidden, restricted or from deleted accounts are not served and are not in the file.

Which format is best for research?

JSON also records the post URL, the displayed comment count and the fetch time. CSV and Excel are easier to open for quick checks.

Can I identify people from the export?

Usernames and user IDs are in the file, so treat it as personal data and apply your ethics board’s rules. Pseudonymise early.

Can I get historical versions of a post’s comments?

No. An export is a snapshot at the time of the fetch. To study change over time, export on a schedule and keep each file.

Is this legal advice?

No. These are method notes. Check your institution’s rules and the law that applies to you.